Anonymous computable objects

The descriptions of real-world datasets are extensive, by consequence it is necessary to allow the definition of user defined conditional and unconditional probability measures so that it is possible to define new computable objects.

A computable object defined with a user defined probability measure is called an anonymous computable object and a user defined probability measure function is called an external probability function. As well as the external functions, introduced in the previous section, the external probability function constructor accepts two kinds of argument, either a C++ function or an object class method with the following prototypes:

plProbValue user_function(const plValues &input_values)

plProbValue object_class::class_method(const plValues &input_values)

The last implies that you will write an object class containing a method corresponding to the probability measure function. Once the external function is constructed, it is used to define an anonymous computable object.

In summary to define a new computable object you have to:

1. Write a C++ function or an object class.
2. If you coded the function by an object class method, construct an instance of the object class.
3. Construct an external probability function with the C++ function or the class method.
4. Construct the anonymous computable object with the external probability function.

The anonymous computable objects are divided in: anonymous kernels and anonymous conditional kernels and are described in the following sections.

• Anonymous kernels
• Anonymous conditional kernels